Evaluation of News Search Engines Based On Information Retrieval Models

News search engines are the exclusive search services for users’ news intake. Providing relevant query to a news search engine, the user gets back a single news result page consisting of various news articles aggregated from thousands of online news sources available on the World Wide Web. The availability and use of major news search engines like Bing news, Google news and Newslookup demand retrieval effectiveness evaluation of these search systems. In this paper, core retrieval models, namely, vector space model, Okapi BM25 and latent semantic indexing are used to evaluate retrieval effectiveness of news search engines for relevance effectiveness evaluation considering these models separately. Further, Monte-Carlo cross-entropy based rank aggregation technique is used to do more comprehensive relevance effectiveness evaluation by aggregating three individual rankings. Experimental results denote Google news’s performance to be better than the other two search engines.